Gestational Diabetes and Pregnancy Outcomes
Gestational Diabetes and Pregnancy Outcomes
Types of Study Cohort studies (prospective or retrospective) were considered for inclusion in this systematic review if they provided sufficient information to estimate the associations of the WHO and/or the IADPSG criteria with related perinatal and maternal outcomes.
To avoid selection bias, we included only studies that applied the OGTT universally to all participants. We therefore excluded studies applying the OGTT only in women with certain clinical risk factors (such as family history, obesity, previous GDM) or in those positive in pre-OGTT glucose screening (with, for example, a 50 g challenge test and/or a fasting glucose). We also excluded studies that did not distinguish pre-gestational diabetes mellitus from GDM, those not allowing the distinction between treated and untreated groups, and those not reporting outcomes for women classified as having a normal OGTT.
Types of Participants We accepted studies which included women of any race, parity, age, body weight or other socio-demographic characteristics.
Types of Diagnostic Tests Only studies based on a 2 hour 75 g OGTT performed during the 2 or the 3 trimesters were included, and only if they provided results for a diagnosis based on at least the 2 h post-load glucose. Studies based on capillary glucose measurements were included.
Types of Outcome Measures We decided to analyze, as perinatal outcomes, large for gestational age births and macrosomia (as defined by the authors), as well as perinatal mortality (fetal death and early neonatal death). Regarding maternal outcomes, we chose to analyze cesarean delivery and preeclampsia according to individual study definitions.
The search strategy used the following general terms, adapted to each database: "gestational diabetes" or "glucoseintolerance" and the appropriate terms for each of the maternal and perinatal adverse outcomes specified above. Specific terms used for the electronic search are detailed in the Additional file 1: Description of the electronic search strategy used to perform the literature search.
We searched ten electronic databases (MEDLINE, EMBASE, LILACS, the Cochrane Library (CENTRAL), CINHAL, WHO-Afro library, IMSEAR, EMCAT, IMEMR and WPRIM) for articles published from inception up to March 15, 2011. No language or country restrictions were applied. We also searched for additional studies from classical review articles. The reference lists of all articles selected for full text reading were reviewed for additional potentially eligible studies.
Selection of Studies All citations identified were entered into an electronic database and duplicates were deleted. Initially, two investigators independently screened the titles and abstracts of potentially relevant studies for eligibility. When the information was not sufficient to determine if the article was eligible for inclusion, the article's full text was obtained for further evaluation. Discrepancies were discussed until consensus was reached.
Data Extraction and Management Two independent investigators reviewed the eligible studies and extracted data using a standardized form prepared for this review. Disagreements were discussed and resolved in a consensus meeting. When raw quantitative data were not reported, approximate values were obtained from the figures or calculated from percentages.
Assessment of Methodological Quality The methodological quality of the included studies was assessed by examining factors which might affect the strength of the association between glucose levels and outcomes. In particular, the following factors were assessed in each study: i) adequate selection of participants: consecutive recruitment from prenatal clinics; ii) adequate standardization of the glucose tolerance test (pre-analytic factors such as anhydrous glucose, plasma immediately separated or kept with glycolytic inhibitors and kept refrigerated until centrifugation; and analytic factors such as enzymatic method of measurement and laboratory quality control); iii) adequate report of losses to follow up and; iv) medical staff blinded to OGTT results.
Data Synthesis Data for the WHO and the IADPSG criteria were aggregated and presented as relative risk (RR) with 95% confidence interval (CI). Meta-analysis data were combined using random-effect models, with restricted maximum-likelihood (REML) estimation. The statistical analysis was performed using the R version 2.11.1 software, package metafor version 1.6–0. As our aim was to investigate diagnostic criteria based on their capacity to predict GDM-related outcomes for classification purposes rather than for etiological investigation, all statistical analyses were crude, without adjustment for potential confounders.
Assessment of Heterogeneity Overall results were calculated based on the random effects model. We assessed heterogeneity using the Cochrane's χ statistics with a significance level of 0.10. Inconsistency indexes (I) were also calculated, and a value greater than 50% was considered an indicator of high inconsistency between studies.
Sensitivity Analysis and Assessment of Publication Bias We did sensitivity analyses in order to examine the influence of the HAPO study and Brazilian Study of Gestational Diabetes (EBDG) on the magnitude and consistency of associations with outcomes. In addition to REML, we also aggregated data with other variance estimators (Maximum Likelihood, Empirical Bayes, Sidik-Jonkman and DerSimonean and Laird) and with a fixed effect model in order to assess the robustness of the model.
Publication bias was tested using a funnel plot and Egger's test based on weighted regression.
The full database for the EBDG study was available which allowed analysis for both criteria for all outcomes. The EBDG study was approved by local institutional review boards and informed consent was obtained from all participants. Data from the other studies were obtained from published articles cited in the list of references.
Methods
Criteria for Considering Studies for This Review
Types of Study Cohort studies (prospective or retrospective) were considered for inclusion in this systematic review if they provided sufficient information to estimate the associations of the WHO and/or the IADPSG criteria with related perinatal and maternal outcomes.
To avoid selection bias, we included only studies that applied the OGTT universally to all participants. We therefore excluded studies applying the OGTT only in women with certain clinical risk factors (such as family history, obesity, previous GDM) or in those positive in pre-OGTT glucose screening (with, for example, a 50 g challenge test and/or a fasting glucose). We also excluded studies that did not distinguish pre-gestational diabetes mellitus from GDM, those not allowing the distinction between treated and untreated groups, and those not reporting outcomes for women classified as having a normal OGTT.
Types of Participants We accepted studies which included women of any race, parity, age, body weight or other socio-demographic characteristics.
Types of Diagnostic Tests Only studies based on a 2 hour 75 g OGTT performed during the 2 or the 3 trimesters were included, and only if they provided results for a diagnosis based on at least the 2 h post-load glucose. Studies based on capillary glucose measurements were included.
Types of Outcome Measures We decided to analyze, as perinatal outcomes, large for gestational age births and macrosomia (as defined by the authors), as well as perinatal mortality (fetal death and early neonatal death). Regarding maternal outcomes, we chose to analyze cesarean delivery and preeclampsia according to individual study definitions.
Search Methods for the Identification of Studies
The search strategy used the following general terms, adapted to each database: "gestational diabetes" or "glucoseintolerance" and the appropriate terms for each of the maternal and perinatal adverse outcomes specified above. Specific terms used for the electronic search are detailed in the Additional file 1: Description of the electronic search strategy used to perform the literature search.
We searched ten electronic databases (MEDLINE, EMBASE, LILACS, the Cochrane Library (CENTRAL), CINHAL, WHO-Afro library, IMSEAR, EMCAT, IMEMR and WPRIM) for articles published from inception up to March 15, 2011. No language or country restrictions were applied. We also searched for additional studies from classical review articles. The reference lists of all articles selected for full text reading were reviewed for additional potentially eligible studies.
Data Collection and Analysis
Selection of Studies All citations identified were entered into an electronic database and duplicates were deleted. Initially, two investigators independently screened the titles and abstracts of potentially relevant studies for eligibility. When the information was not sufficient to determine if the article was eligible for inclusion, the article's full text was obtained for further evaluation. Discrepancies were discussed until consensus was reached.
Data Extraction and Management Two independent investigators reviewed the eligible studies and extracted data using a standardized form prepared for this review. Disagreements were discussed and resolved in a consensus meeting. When raw quantitative data were not reported, approximate values were obtained from the figures or calculated from percentages.
Assessment of Methodological Quality The methodological quality of the included studies was assessed by examining factors which might affect the strength of the association between glucose levels and outcomes. In particular, the following factors were assessed in each study: i) adequate selection of participants: consecutive recruitment from prenatal clinics; ii) adequate standardization of the glucose tolerance test (pre-analytic factors such as anhydrous glucose, plasma immediately separated or kept with glycolytic inhibitors and kept refrigerated until centrifugation; and analytic factors such as enzymatic method of measurement and laboratory quality control); iii) adequate report of losses to follow up and; iv) medical staff blinded to OGTT results.
Data Synthesis Data for the WHO and the IADPSG criteria were aggregated and presented as relative risk (RR) with 95% confidence interval (CI). Meta-analysis data were combined using random-effect models, with restricted maximum-likelihood (REML) estimation. The statistical analysis was performed using the R version 2.11.1 software, package metafor version 1.6–0. As our aim was to investigate diagnostic criteria based on their capacity to predict GDM-related outcomes for classification purposes rather than for etiological investigation, all statistical analyses were crude, without adjustment for potential confounders.
Assessment of Heterogeneity Overall results were calculated based on the random effects model. We assessed heterogeneity using the Cochrane's χ statistics with a significance level of 0.10. Inconsistency indexes (I) were also calculated, and a value greater than 50% was considered an indicator of high inconsistency between studies.
Sensitivity Analysis and Assessment of Publication Bias We did sensitivity analyses in order to examine the influence of the HAPO study and Brazilian Study of Gestational Diabetes (EBDG) on the magnitude and consistency of associations with outcomes. In addition to REML, we also aggregated data with other variance estimators (Maximum Likelihood, Empirical Bayes, Sidik-Jonkman and DerSimonean and Laird) and with a fixed effect model in order to assess the robustness of the model.
Publication bias was tested using a funnel plot and Egger's test based on weighted regression.
The full database for the EBDG study was available which allowed analysis for both criteria for all outcomes. The EBDG study was approved by local institutional review boards and informed consent was obtained from all participants. Data from the other studies were obtained from published articles cited in the list of references.
Source...